The transition from adolescence to adulthood is characterized by rapid increases in substance use, including nicotine, alcohol, and illegal drugs. For most people, substance use increases during adolescence, peaks between ages 18 to 25, then decreases substantially after age 25. However, many people do not follow this normal progression, and some show more serious patterns of substance use-they start using at a younger age, use multiple substances, become addicted, or fail to show the normal decreases in substance use after their mid-twenties. Therefore, there are multiple trajectories of substance use that people follow as they age, some of which lead to substantially worse life outcomes. The current project focuses on these high-risk trajectories. To increase the effectiveness of intervention efforts for substance users, it is important to (1) understand differences between high-risk and normal trajectories, (2) identify people who are starting to follow these high-risk trajectories at an early age, and (3) pinpoint features of these people's lives that can be modified with treatment (e.g., anxiety symptoms, delinquent friends, parent use of substances, etc.). The aims of my proposed research plan closely follow these three priorities. Specifically, I will use longitudinal twin data to (1) identify different trajectories of substance use across adolescence and young adulthood (ages 11-29), (2) use sex and childhood personality to predict which of these trajectories each person is likely to follow, and (3) use a co-twin control design to identify a pathway from early childhood risk variables to adult substance use outcomes. Co-twin control designs can establish the reasons why variables such as personality and substance are correlated, providing a stronger basis for causal inference than traditional correlational approaches. This is primarily because the design accounts for the effects of genetic overlap and shared environment, leaving only the effects of unique environment. Because twins are matched on genotype and family background, differences between twins provide causal evidence. For example, finding different substance use outcomes within identical twin pairs that differ, correspondingly, in neuroticism would give greater inferential certainty about causality than the observation of a simple association between substance use and neuroticism in unrelated individuals. This proposal has three key strengths. First, we will work with a large and representative twin sample to achieve these aims. Second, the team of mentors is very well qualified to conduct this research, in terms of both substantive and methodological expertise. Third, the research will be conducted by the Minnesota Center for Twin and Family Research, which has an excellent track record of successful NIH-funded research with community twin samples. Taken together, the proposed projects are part of a training experience and research course meant to chart the causal progression from premorbid risk factors to substance use outcomes, identify the modifiable factors associated with substance use, and examine the role of these factors within various trajectories of use.